`multi_predict._multnet()` doesn't pull the right default `penalty`

[DavisVaughan]

We are using object$lambda rather than object$fit$lambda and that causes:

library(parsnip)
data("lending_club")

multi_reg <- multinom_reg()
multi_reg_glmnet <- multi_reg %>% set_engine("glmnet")
multi_reg_fit <- fit(multi_reg_glmnet, verification_status ~ annual_inc + sub_grade, data = lending_club)

multi_reg_fit %>%
  multi_predict(new_data = lending_club, type = "prob", penalty = 0.01)
#> # A tibble: 9,857 x 1
#>    .pred           
#>    <list>          
#>  1 <tibble [1 × 4]>
#>  2 <tibble [1 × 4]>
#>  3 <tibble [1 × 4]>
#>  4 <tibble [1 × 4]>
#>  5 <tibble [1 × 4]>
#>  6 <tibble [1 × 4]>
#>  7 <tibble [1 × 4]>
#>  8 <tibble [1 × 4]>
#>  9 <tibble [1 × 4]>
#> 10 <tibble [1 × 4]>
#> # … with 9,847 more rows

multi_reg_fit %>%
  multi_predict(new_data = lending_club, type = "prob")
#> Error in arrange_impl(.data, dots): cannot arrange column of class 'NULL' at position 2

This is the line that needs to be updated:
https://github.com/tidymodels/parsnip/blob/master/R/logistic_reg.R#L286

[github-actions]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.